How to resume the training of an already stopped model changing a parameter of the model?

Question

I have a build model function which change through a parameter the nature of a lambda layer inside the model itself:

def get_model2(mask_kind):
   epochs = 20